This article contains frequently asked questioned about Apache Kafka on Heroku Maintenance.
Why is this maintenance happening?
Apache Kafka on Heroku is a managed Kafka offering, and one of the largest values provided is providing patches for security and feature updates. Heroku monitors for and proactively patches security vulnerabilities as part of its Apache Kafka on Heroku offering.
How can I protect my app against downtime and errors during maintenance?
Please review our Robust Usage guidelines. This will not only protect you against errors during maintenance, but also during rare but possible Kafka node outages.
How long will maintenance take?
Maintenance on a single cluster should only take ~10-15 minutes per service.
standard-Xplan: ~2 hours
extended-Xplan: ~3.5 hours
As a customer's cluster has maintenance performed, Kafka clients may see small amounts of errors as partitions move between brokers.
Why is my Kafka addon "recovering"?
When Kafka or Zookeeper services are being replaced, the
heroku kafka:info command will show the status as "recovering".
=== KAFKA_URL Plan: heroku-kafka:standard-0 Status: recovering ...
This status should only last for the duration of the maintenance.
How do I resolve
The fastest and most simple way to recover from this error is to restart your consumer and producer dynos.
For more information on preventing this error, please refer to the Partition Leader section of our Robust Usage document.